Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from Original research Fluorene exposure among PAH-­exposed workers is associated with epigenetic markers related to lung cancer Ayman Alhamdow ‍ ‍ ,1 Yona J Essig,2 Annette M Krais,2 Per Gustavsson ‍ ‍ ,1,3 Håkan Tinnerberg,4 Christian H Lindh,2 Jessika Hagberg,5,6 Pål Graff ‍ ‍ ,7 Maria Albin,1,2,3 Karin Broberg ‍ ‍ 1,2

►► Additional material is Abstract Key messages published online only. To view Objectives Exposure to high-molecular-­ ­weight please visit the journal online polycyclic aromatic hydrocarbons (PAHs) may cause (http://dx.​ ​doi.org/​ ​10.1136/​ ​ What is already known about this subject? cancer in chimney sweeps and creosote-exposed­ workers, oemed-2020-​ ​106413). ► Occupational exposure to high-­molecular-­ however, knowledge about exposure to low-­molecular-­ ► weight polycyclic aromatic hydrocarbons (PAHs) For numbered affiliations see weight PAHs in relation to cancer risk is limited. In this has been widely studied in relation to cancer end of article. study, we aimed to investigate occupational exposure biomarkers, however, less is known about to the low-molecular­ -­weight PAHs phenanthrene and Correspondence to exposure to low-­molecular-­weight PAHs and fluorene in relation to different cancer biomarkers. Dr Karin Broberg, Institute their carcinogenicity. of Environmental Medicine, Methods We recruited 151 chimney sweeps, 19 ►► Workers exposed to PAHs are at a higher risk of Karolinska Institutet, Stockholm creosote-­exposed workers and 152 unexposed workers cancer. SE-171 76, Sweden; (controls), all men. We measured monohydroxylated ► Hypomethylation of F2RL3 and AHRR has been karin.​ ​broberg@ki.​ ​se metabolites of phenanthrene and fluorene in urine ► shown to be marker for lung cancer risk. AA and YJE contributed equally. using liquid chromatography coupled to tandem mass spectrometry. We measured, in peripheral blood, the What are the new findings? Received 9 January 2020 cancer biomarkers telomere length and mitochondrial ►► Chimney sweeps and creosote-­exposed workers Revised 2 April 2020 DNA copy number using quantitative PCR; and DNA Accepted 5 April 2020 were exposed to the low-molecular­ -weight­ methylation of F2RL3 and AHRR using pyrosequencing. PAHs fluorene and phenanthrene. Results Median PAH metabolite concentrations ►► Fluorene exposure was associated with the were higher among chimney sweeps (up to 3 times) hypomethylation of F2RL3 and AHRR. and creosote-exposed­ workers (up to 353 times),

compared with controls (p<0.001; adjusted for age How might this impact policy or clinical prac- http://oem.bmj.com/ and smoking). ∑OH-­fluorene (sum of 2-­hydroxyfluorene tice in the foreseeable future? and 3-hydro­ xyfluorene) showed inverse associations ►► This study highlights the importance of with percentage DNA methylation of F2RL3 and AHRR measuring not only high-molecular­ -­weight PAHs in chimney sweeps (B (95% CI)=–2.7 (–3.9 to –1.5) but also the low-­molecular-­weight PAHs when for F2RL3_cg03636183, and –7.1 (–9.6 to –4.7) for assessing occupational PAH exposure. AHRR_cg05575921: adjusted for age and smoking), but ►► Further, it indicates that even exposure to

not in creosote-­exposed workers. In addition, ∑OH-­ low-molecular­ -weight­ PAHs might have a on September 27, 2021 by guest. Protected copyright. fluorene showed a 42% mediation effect on the inverse carcinogenic effect. If this is the case, then the association between being a chimney sweep and DNA number of workers under risk is much larger methylation of AHRR CpG2. than currently assumed, as many workers are Conclusions Chimney sweeps and creosote-exposed­ exposed to low-molecular­ -weight­ PAHs only. workers were occupationally exposed to low-­molecular-­ weight PAHs. Increasing fluorene exposure, among chimney sweeps, was associated with lower DNA Finland, Denmark and Iceland).2 For instance, methylation of F2RL3 and AHRR, markers for increased standardised incidence ratio (SIR) and 95% CI for lung cancer risk. These findings warrant further chimney sweeps in all countries were SIR 2.41 investigation of fluorene exposure and toxicity. (95% CI: 1.40 to 3.85) for pharynx cancer, and SIR 1.86 (95% CI: 1.19 to 2.76) for oesophageal © Author(s) (or their cancer.2 Moreover, high risk for cancer was seen in employer(s)) 2020. Re-­use permitted under CC BY. a cohort of chimney sweeps (n=6320) from Sweden Published by BMJ. Introduction (eg, SIR=1.30 (95% CI: 1.21 to 1.39) for total Scrotal cancer among chimney sweeps was the cancer; and SIR=2.08 (95% CI: 1.19 to 3.38) for To cite: Alhamdow A, first cancer form linked to chemical exposure.1 oesophageal cancer).3 The same study found that Essig YJ, Krais AM, et al. Occup Environ Med Epub Even today, the risk of occupational cancer among the number of working years as a chimney sweep (a ahead of print: [please include chimney sweeps is very high, based on analysis of proxy for soot exposure) was positively associated Day Month Year]. doi:10.1136/ 53 occupational groups (n=15 million individ- with cancer incidence.3 The major carcinogenic oemed-2020-106413 uals) from the Nordic countries (Sweden, Norway, exposure for the chimney sweeps is considered to

Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 1 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from be polycyclic aromatic hydrocarbons (PAHs)—the main constit- for controls, and on Thursdays for creosote-­exposed workers, uent of soot.4 5 The International Agency for Research on Cancer during which trained nurses collected questionnaires, sampled (IARC) has classified PAHs as human carcinogens (group 1: eg, blood and urine, and measured weight and height. The ques- benzo[a]pyrene; BaP), probable human carcinogens (group 2A: tionnaire included questions about age, chronic diseases, level eg, dibenz[a,h]anthracene), possible human carcinogens (group of education, tobacco smoking history, intake of dietary items 2B: eg, benz[a]anthracene; BaA) or not classifiable as to their including vegetables, fruits and fish, use of chewing tobacco carcinogenicity to humans (group 3: eg, pyrene, phenanthrene (snus), passive smoking, physical activity, residential history, (Phe) and fluorene (Flu)).6 Creosote-­impregnating workers is employment history and exposure from hobbies. The question- another occupational group exposed to PAHs with an elevated naire of chimney sweeps further explored the extent of workers’ risk for cancer.6 7 Workers in this occupation are exposed to engagement in different sweeping tasks (eg, soot sweeping in PAHs from creosote oil, which is classified as a probable human private houses as well as in industrial buildings) during different carcinogen.6 time periods (eg, during the past 12 months). We previously demonstrated, by measuring four PAH metab- Postshift spot urine and EDTA blood samples were collected olites in urine, that currently working chimney sweeps and from participants and transported at room temperature to the creosote-­exposed workers in Sweden were exposed to the Division of Occupational and Environmental Medicine, Lund PAHs pyrene and phenanthrene, BaA and BaP (BaP was only University for storage at −20°C. Five chimney sweeps and four found in chimney sweeps).4 8 Notably, the highest PAH expo- controls refrained from donating blood samples, whereas three sure, measured as urinary PAH metabolites, found in creosote-­ chimney sweeps and one control could not urinate. All creosote-­ exposed workers was phenanthrene—a low-­molecular-­weight exposed workers donated blood and urine samples. Written PAH (median 2-­OH-­phenanthrene concentrations in urine were informed consent was obtained from all study participants. The 29.6 vs 0.14 µg/g creatinine for creosote-­exposed workers and study was carried out in accordance with the ethical guidelines controls, respectively). Phenanthrene and other low-­molecular-­ of the Declaration of Helsinki 1964 and its later amendments weight PAHs, such as fluorene, might thus be interesting and was approved by the Regional Ethics Committee in Lund, targets for biomonitoring studies of PAH-­exposed workers. We Sweden and the Regional Ethics Committee in Uppsala, Sweden. also showed that both chimney sweeps and creosote-­exposed workers had lower DNA methylation of the genes F2RL3 and PAH metabolites in urine AHRR—prospective biomarkers for lung cancer.8 9 We did not We measured urinary monohydroxylated metabolites of phenan- find any associations with relative telomere length (TL) or threne (1-­OH-­Phe, 2-­OH-­Phe, 3-­OH-­Phe and 4-­OH-­Phe) and mitochondrial DNA copy number (mtDNAcn), two biomarkers fluorene (2-­OH-Flu­ and 3-­OH-Flu)­ by liquid chromatography previously linked to cancer risk,10 11 nor did we observe associ- coupled to tandem mass spectrometry (LC-MS/MS).­ Details of ations between the four measured PAH metabolites and DNA the method were described previously.4 Urine samples (0.2 mL) methylation levels.8 The latter observation may be due to differ- were pipetted into a 96-­well plate and hydrolysed by adding ences in half-lives­ between the urinary PAH metabolites and 0.1 mL of ammonium acetate (pH 6.5) and 0.01 mL of β-glucuro- DNA methylation, or alternatively, that the four PAHs analysed nidase (Escherichia coli). The solution was incubated for 30 min do not cause epigenetic changes. It therefore remains to clarify at 37°C. Afterwards, 0.025 mL of a 50:50 (v/v) water/acetonitrile if PAHs present in soot and creosote oil can cause the cancer-­ solution and 0.025 mL of deuterium-labelled­ internal standards related DNA methylation. This will increase our knowledge of http://oem.bmj.com/ (D -1-­OH-­Phe, D -2-­OH-­Phe, D -4-­OH-­Phe, D -2-­OH-­Flu; occupational PAH exposure and related risk of cancer. 9 9 9 9 Toronto Research Chemicals, North York, Ontario, Canada) We aimed in this study to (i) further characterise the expo- were added (final concentrations 5 ng/mL). Plates were centri- sure to phenanthrene, (ii) investigate exposure to fluorene and fuged for 10 min at 3000 rpm, thereafter, 3 µL of each sample (iii) evaluate the associations of phenanthrene and fluorene were injected onto the column and analysed using a Shimadzu exposures with early biomarkers of cancer (TL, mtDNAcn and UFLC system (Shimadzu Corporation, Kyoto, Japan), coupled DNA methylation), among chimney sweeps, creosote-exposed­ to QTRAP6500+ (AB Sciex, Foster City, California, USA) and workers and controls. on September 27, 2021 by guest. Protected copyright. equipped with a Turbolon Spray source (AB Sciex). The separa- tion was achieved using a Kinetex C18 column (2.6 µm, 100 Å, Methods 2.1 mm i.d. × 100 mm, Phenomenex, Torrance, California, Study design and participants USA). The mobile phase was water (A) and acetonitrile (B). The gradient was linearly increased from 5% B to 75% B within This is a cross-­sectional study based on the following occupational 6 min, then to 95% B within 10 s and held there for 1 min. 2-­OH-­ groups: chimney sweeps (n=151), control workers occupation- Phe and 3-­OH-Phe­ as well as 2-­OH-Flu­ and 3-­OH-Flu­ could ally unexposed to PAHs (n=152) and a small group of creosote-­ not be separated and were therefore analysed as single peaks exposed workers (n=19); all men and working in Sweden. The (∑2-­OH-Phe,­ 3-­OH-Phe­ and ∑OH-­Flu, respectively). Samples latter group was included due to their known high exposure to were prepared and run in duplicates and the average values were PAHs from the creosote oil, which gives us the possibility to used for further analysis. Limit of detection (LOD) values were evaluate both moderate (chimney sweeps) and high (creosote-­ calculated from water blanks. All PAH metabolite concentrations exposed workers) PAH exposure. Detailed information about were adjusted to urinary creatinine values ( g/g crea.). study groups, including recruitment, work tasks and sampling, µ have previously been described.4 8 In brief, the recruitment process took place during 2010–2015 for chimney sweeps and DNA methylation of F2RL3 and AHRR controls and during 2013–2016 for creosote-­exposed workers. DNA methylation data used in this study were reported in a The company response rate was 87% for chimney sweeps, previous publication.8 Hypomethylation of the CpG cg03636183 58% for controls and 100% for creosote-­exposed workers.4 (Illumina 450K bead chip ID) in exon 2 of F2RL3 (encoding the Company visits were performed in the afternoons of Wednes- coagulation factor II (thrombin) receptor-like­ 3) and cg05575921 days or Thursdays for chimney sweeps, throughout the week in intron 3 of AHRR (encoding the Aryl Hydrocarbon Receptor

2 Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from

Repressor) is strongly associated with tobacco smoking and has Unadjusted and multivariable linear regression models in prospective studies been associated with increased lung cancer were fit to evaluate associations between the PAH metabolite risk.9 12 We analysed DNA methylation of two CpGs in F2RL3 and concentrations and cancer biomarkers in study groups: model three CpGs in AHRR, including cg03636183 and cg05575921. 1, unadjusted; and model 2, adjusted for age and smoking (a A detailed description of the method is previously provided.8 In priori confounders). Unstandardised beta estimate (B) and brief, DNA samples (500 ng), extracted from peripheral blood, 95% CI were presented for each association. To disentangle were bisulfite-treated­ using EZ-96 DNA Methylation-­Gold Kit the effect of smoking from the effect of occupational PAH (Zymo Research, Irvine, California, USA). All samples were exposure on the association between ∑OH-­Flu and DNA randomised and DNA methylation controls (0% and 100% methylation, an additional model was fit (model 3) among methylation; Qiagen GmbH, Hilden, Germany) were added to smoking chimney sweeps (n=27) and smoking controls (n=25) 96-well­ plates. Bisulfite-treated­ DNA sequences of interest were adjusting for age and pack-years.­ Creosote-­exposed workers amplified using the thermal cycler (SimpliAmp; Thermo Fisher were not considered in model 3 due to the limited number of Scientific, Carlsbad, California, USA) and DNA methylation smokers (n=3). We adjusted for multiple comparisons (false was measured using PyroMark Q96 ID (Qiagen). All PCR and discovery rate <0.05) in models 1 and 2. Further adjustment pyrosequencing chemicals and reagents were purchased from was evaluated for use of snus, physical activity, intake of fish, Qiagen except for Streptavidin Sepharose High Performance level of education, current residential area, exposure to smoke (GE Healthcare, Uppsala, Sweden). We randomly selected and from a hobby and passive smoking. In addition, the associ- reran 10% of the samples to calculate the coefficient of variation ations between ∑OH-Flu­ and cancer biomarkers were eval- for each CpG (ranged between 2.5% and 4.1%). uated among never smoking chimney sweeps and controls adjusting for age. For all linear regression models, key model assumptions were evaluated. Relative TL and mtDNAcn Finally, we analysed whether PAH metabolites are medi- Data of relative TL and relative mtDNAcn were reported previ- 8 ating the observed chimney sweeps’ lower DNA methylation ously. Measurement was performed by quantitative PCR (qPCR) of AHRR and F2RL3 (7), that is, chimney sweeping PAH using DNA samples extracted from peripheral blood. The copy → exposure (PAH metabolite) lower DNA methylation. Medi- number of telomere repeats (T), copy number of mitochondrial → ation percentage was calculated based on the formula (medi- DNA (M) and copy number of a single copy gene (HBB: haemo- ation % = [(B −B)/B ] ×100), where B is the unstandardised globin beta) (S) were calculated from independent qPCR runs 0 0 0 estimate of the independent variable ‘study group’ in the base using a standard curve. Thereafter, relative TL was calculated as model (eg, DNA methylation=intercept + B × study group the ratio T/S and relative mtDNAcn as the ratio M/S. All study 0 + B × age), and B is the unstandardised estimate of ‘study samples, negative controls and standard curve DNA were run in 1 group’ after introducing the variable (mediator) PAH metabo- triplicates (SD <0.2 was accepted for C values of the triplicates) t lite (such as ∑OH-Flu)­ to the model (DNA methylation=inter- and 10% of the samples were randomly selected and rerun to cept + B × study group + B × age + B × PAH metabolite).13 ensure quality control. The coefficient of variation was 7.3% 1 2 for relative TL and 14.4% for relative mtDNAcn. The coeffi- We only evaluated mediation for CpGs showing p<0.2 in the cient of determination (R2) for all qPCR runs of the telomere, base model. HBB and mtDNAcn was >0.99. Further information about PCR All statistical analyses were performed with SPSS statistical http://oem.bmj.com/ conditions, standard curve and primer sequences is described software V.25.0. A P value <0.05 was considered for statistical elsewhere.8 significance.

Statistical analysis Results Median, minimum, and maximum values were calculated for Basic characteristics of study groups, as well as levels of cancer the continuous variables (age, body mass index (BMI), cancer biomarkers, are described in table 1. Controls had slightly

biomarkers, and PAH metabolites). Frequencies (percentages) higher BMI compared with chimney sweeps (BMI data were on September 27, 2021 by guest. Protected copyright. were calculated for categorical variables (eg, use of snus and not available for creosote-­exposed workers). The use of snus cigarette smoking). Differences between study groups were and current residential area differed between study groups. evaluated by the Kruskal-­Wallis test for continuous variables As reported in our previous publication, TL and mtDNAcn and by the Fisher’s exact test for categorical variables. To eval- did not differ between study groups.8 Hypomethylation of uate PAH metabolite concentration differences between study F2RL3 (CpG1 and CpG2_cg03636183) and AHRR (CpG1, groups, age-­adjusted and smoking-­adjusted linear regression CpG2, and CpG3_cg05575921) was observed among chimney models (analysis of all participants) or age-adjusted­ models sweeps and creosote-­exposed workers compared with controls (analysis in each smoking stratum) were used. Intercorrela- (table 1).8 tions of PAH metabolites, including high-­molecular-weight­ Chimney sweeps had up to three times higher median metabolites that were analysed in a previous publication,8 concentrations of all urinary PAH metabolites measured were evaluated by Spearman’s correlations and presented for (1-­OH-­Phe; 0.23 µg/g crea., ∑2-­OH-­Phe, 3-­OH-­Phe; former/never smoking chimney sweeps and controls. The sum 0.37 µg/g crea., 4-­OH-­Phe; 27 ng/g crea., and ∑OH-Flu;­ of all metabolites of phenanthrene (∑OH-­Phe=1-­OH-­Phe + 0.32 µg/g crea.) than the controls (0.14 µg/g crea., 0.11 µg/g ∑2-­OH-Phe,3-­ ­OH-Phe+4-­ ­OH-Phe:­ µg/g crea.) was calculated crea., 15 ng/g crea. and 0.15 µg/g crea., respectively; in order to compare with other studies. Differences in urinary p<0.001; all participants, adjusting for age and smoking; PAH metabolite concentrations for chimney sweeps who spent table 2). Creosote-­exposed workers had very high concentra- ≥50% of their time doing soot sweeping in the past 12 months tions of all urinary PAH metabolites, for example, up to 353 versus those who spent <50% (excluding current smokers) times higher median concentrations of ∑OH-­Flu (53 µg/g were presented as boxplots and evaluated by Mann-­Whitney crea.) as compared with the controls (p<0.001; all partici- U test. pants, adjusted for age and smoking). The concentrations of

Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 3 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from

Table 1 Basic characteristics of study groups Chimney sweeps Creosote-­exposed Covariates Controls (n=152)* (n=151)* workers (n=19)* P value† Age (years) 43 (20–63) 43 (19–66) 32 (22–58) 0.492 BMI (kg/m2) 27 (20–45) 26 (19–37) – 0.019 Relative telomere length 0.57 (0.35–1.0) 0.56 (0.36–0.97) 0.57 (0.33–0.78) 0.836 Relative mitochondrial DNA copy number 0.87 (0.54–1.7) 0.83 (0.51–1.41) 0.95 (0.43–1.34) 0.162 F2RL3_CpG1 89.5 (58–100) 88.8 (55–99) 87.2 (65–96) 0.019 F2RL3_CpG2 (cg03636183) 76.7 (48–86) 76.6 (45–83) 73.3 (50–78) 0.011 AHRR_CpG1 75.0 (25–88) 73.0 (21–90) 74.1 (31–83) 0.087 AHRR_CpG2 69.1 (23–80) 66.5 (20–77) 68.7 (31–82) 0.004 AHRR_CpG3 (cg05575921) 90.0 (35–100) 88.1 (29–100) 84.9 (37–94) 0.006 Cigarette smoking Current smoker 25 (16.4) 27 (17.9) 3 (15.8) 0.915 Party/ former smoker 44 (28.9) 49 (32.5) 6 (31.6) Never smoker 83 (54.6) 74 (49.0) 9 (47.4) Snus (yes) 29 (19.1) 52 (34.4) 6 (31.6) 0.004 Passive smoking (yes) 24 (15.8) 28 (18.5) 5 (26.3) 0.384 Exposure from hobby (yes) 6 (3.9) 9 (6.0) 2 (10.5) 0.268 Vegetables‡ (≥5 times per week) 88 (57.9) 96 (63.6) 11 (57.9) 0.609 Fruits§ (≥5 times per week) 88 (57.9) 75 (49.7) 10 (52.6) 0.351 Fish¶ (≥1 per week) 70 (46.1) 73 (48.3) 9 (47.4) 0.884 Physical activity**(high) 62 (40.8) 73 (48.3) 7 (36.8) 0.370 Education (higher than high school) 31 (20.4) 24 (15.9) 0 (0)†† 0.100 Current residency (big city) 70 (46.1) 52 (34.4) 11 (57.9)†† 0.006 Values are median (min–max) for continuous variables or n (%) for categorical variable. The content of this table was previously reported (Alhamdow et al, 2018). *Up to two missing cases for some of the variables. †P value of Kruskal-­Wallis test for continuous variables and Fisher’s exact test for categorical variables. ‡Intake of all kinds of vegetables, legumes and root vegetables (fresh, frozen, canned, stewed, juice, soup and so on). §Intake of all kinds of fruits and berries (fresh, frozen, canned, juice, jam and so on). ¶Intake of all kinds of fish. **At least 30 min per week of sudorific activities such as running and badminton during leisure time. ††Missing data for four participants. BMI, body mass index. http://oem.bmj.com/ 4-­OH-­Phe among chimney sweeps and controls were very 0.27), AHRR_ cg05575921, model 2, table 3, adjusted for age, low (unit ng/g crea.). smoking and FDR). Further adjustment for use of snus, phys- There were higher median concentrations of all PAH ical activity, intake of fish, level of education, current residen- metabolites among chimney sweeps across smoking catego- tial area, exposure to smoke from a hobby and passive smoking ries (smokers, party/former smokers and never smokers), did not significantly influence the estimates. Restricting the compared with the controls (table 2). Smokers (both chimney analysis to smokers adjusting for age and pack-­years, ∑OH-Flu­

sweeps and controls) had increased concentrations of ∑OH-­ showed inverse associations with DNA methylation of all five on September 27, 2021 by guest. Protected copyright. Flu in urine compared with the other two smoking categories. analysed CpG sites of F2RL3 and AHRR in chimney sweeps and The phenanthrene metabolites showed a similar but the less controls, but significance was reached in chimney sweeps only, pronounced pattern with smoking categories (table 2). Low-­ and the estimates were about twice as large in chimney sweeps molecular-weight­ PAH metabolites showed stronger inter- (eg, B (95% CI)= –10.0 (–16.0 to –5.1), AHRR_CpG3, model correlations among former/never smoking chimney sweeps 3, table 3) compared with controls (B (95% CI)= –5.5 (–11.0 to

(rS=0.72–0.92) compared with former/never smoking controls 0.10), AHRR_CpG3, model 3, table 3). Analysis among never

(rS=0.40–0.72) (online supplementary table S1). Further, all smoking chimney sweeps and controls showed attenuated effect measured PAH metabolites showed higher concentrations in estimates and there was no clear association between ∑OH-­Flu chimney sweeps who reported high soot sweeping (≥50% and DNA methylation (online supplementary table S2). Metab- of their working time) during the past 12 months compared olites of phenanthrene did not show associations with DNA with those who performed less soot sweeping (<50%) (online methylation (online supplementary table S3). supplementary figure S1; p<0.001, Mann-­Whitney U test). In the mediation analysis, the hypomethylation of AHRR Linear regression analyses showed strong inverse associ- CpG2 was significantly associated with being a chimney sweep

ations between ∑OH-­Flu and DNA methylation of all five in the base model, and the effect estimate (ie, B0) was attenu- analysed CpGs of F2RL3 and AHRR in chimney sweeps (B ated by 42% when ∑OH-­Flu was added to the model (table 4). (95% CI)= –7.1 (–9.6 to –4.7) for AHRR_ cg05575921, model A marked attenuation was also observed for the other CpGs, 2, table 3, adjusted for age, smoking and FDR) and controls and the P values of the effect estimates were considerably (B (95% CI)= –8.2 (–12.0 to –4.6), AHRR_ cg05575921, larger after adding ∑OH-­Flu to the model. For example, B model 2, table 3, adjusted for age, smoking and FDR), but not estimate (95% CI) for AHRR_cg05575921 changed from –2.0 in creosote-­exposed workers (B (95% CI)= –0.062 (–0.4 to (–4.2 to 0.063) to –0.64 (–2.6 to 1.3) (table 4).

4 Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from

Table 2 Median concentrations of urinary PAH metabolites for study groups Controls Chimney sweeps Creosote-­exposed workers PAH metabolite n Median Min Max n* Median Min Max P value† n* Median Min Max P value‡ 1-OH-­ ­Phe (µg/g crea.)  All participants 151 0.14 0.016 2.3 148 0.23 0.032 1.7 <0.001§ 19 19 5.1 61 <0.001§ Smokers 24 0.21 0.074 0.73 27 0.30 0.044 0.67 0.052 3 27 20 29 0.005 Party/former smokers 44 0.15 0.030 0.85 49 0.28 0.053 1.7 <0.001 6 16 5 29 <0.001 Never smokers 83 0.12 0.016 2.3 71 0.18 0.032 1.3 <0.001 9 14 6 55 <0.001 ∑2-,3-OH-­ ­Phe (µg/g crea.)  All participants 151 0.11 0.027 2.7 148 0.37 0.048 4.6 <0.001§ 19 35 11 92 <0.001§ Smokers 24 0.27 0.041 1.3 27 0.50 0.080 2.1 0.004 3 49 41 49 0.005 Party/ former smokers 44 0.11 0.044 0.62 49 0.42 0.055 4.6 <0.001 6 24 11 49 <0.001 Never smokers 83 0.10 0.027 2.7 71 0.31 0.048 3.1 <0.001 9 28 13 92 <0.001 4-OH-­ ­Phe (ng/g crea.)  All participants 151 15 2.6 95 148 27 3.8 304 <0.001§ 19 1051 293 9780 <0.001§ Smokers 24 40 7.6 95 27 46 6.4 152 0.462 3 961 759 1144 0.005 Party/ former smokers 44 16 2.8 92 49 30 3.8 304 <0.001 6 1186 293 7513 <0.001 Never smokers 83 13 2.6 58 71 20 4.4 175 <0.001 9 1051 300 9780 <0.001 ∑OH-­Phe (µg/g crea.)  All participants 151 0.27 0.070 5.1 148 0.64 0.086 6.6 <0.001§ 19 51 19 138 <0.001§ Smokers 24 0.50 0.13 2.1 27 0.88 0.13 2.9 0.010 3 78 62 78 0.005 Party/ former smokers 44 0.28 0.12 1.1 49 0.71 0.12 6.6 <0.001 6 44 19 72 <0.001 Never smokers 83 0.23 0.070 5.1 71 0.54 0.086 4.3 <0.001 9 42 19 138 <0.001 ∑OH-­Flu (µg/g crea.)  All participants 151 0.15 0.042 3.8 148 0.32 0.063 4.3 <0.001§ 19 53 27 128 <0.001§ Smokers 24 1.4 0.21 3.8 27 1.5 0.22 3.4 0.473 3 79 71 118 0.005 Party/ former smokers 44 0.15 0.073 0.82 49 0.39 0.066 4.3 <0.001 6 46 42 70 <0.001 Never smokers 83 0.13 0.042 0.78 71 0.20 0.063 1.9 <0.001 9 38 27 100 <0.001 *There was one participant with missing smoking status. †General linear model comparing controls with chimney sweeps adjusted for age. ‡General linear model comparing controls with creosote-exposed­ workers adjusted for age. §Further adjusted for smoking. ∑OH-­Flu, sum of 2-­hydroxyfluorene and 3-hydroxyfluorene;­ ∑OH-­Phe, sum of 1-­hydroxyphenanthrene, 2-­hydroxyphenanthrene, 3-­hydroxyphenanthrene and 4-­hydroxyphenanthrene; ∑2-,3-­OH-­Phe, sum of 2-­hydroxyphenanthrene and 3-­hydroxyphenanthrene; PAH, polycyclic aromatic hydrocarbon. http://oem.bmj.com/

Discussion phenanthrene and fluorene metabolites were two to three times In this study, we showed that chimney sweeps and particularly higher in postshift compared with preshift samples.17 Further, creosote-exposed­ workers are occupationally exposed to the low-­ workers in the production of fireproof stone were exposed to molecular-weight­ PAHs phenanthrene and fluorene. Moreover, phenanthrene (median 1-OH-­ ­Phe, 2-­OH-­Phe and 4-­OH-­Phe were chimney sweeps showed the hypomethylation of F2RL3 and 1.9, 1.6 and 0.3 µg/g crea., respectively).18 In our study, chimney 9

AHRR—prospective markers for increased risk of lung cancer, sweeps who were soot sweeping ≥50% of their working time had on September 27, 2021 by guest. Protected copyright. with increasing fluorene metabolite concentrations. These results significantly higher phenanthrene and fluorene metabolites in their remained after sensitivity analyses among currently smoking urine compared with chimney sweeps who spent <50% of their chimney sweeps and controls, where potential residual confounding working time sweeping soot (more than twofold increase). In addi- of pack-years­ of smoking could be taken into account. The associa- tion, comparing the exposure levels in our study with those of the tion between fluorene and DNA methylation seems to be fluorene-­ aforementioned occupational groups, it is apparent that the expo- specific as other PAH metabolites, analysed here (metabolites of sure levels of both phenanthrene and fluorene in creosote-­exposed phenanthrene) or in our previous study (pyrene, BaP and BaA),8 workers (median ∑2-­OH-Phe,­ 3-­OH-Phe­ 35 µg/g crea., ∑OH-Flu­ did not show such association. 53 µg/g crea.), but not in chimney sweeps (median ∑2-­OH-Phe,­ Occupational exposure to phenanthrene and fluorene has been 3-­OH-­Phe 0.37 µg/g crea., ∑OH-­Flu 0.32 µg/g crea.), were higher given less attention compared with PAHs such as pyrene (widely than all other occupational groups reported. used as a surrogate for total PAH exposure) and BaP (group-1 Another source of phenanthrene and fluorene exposures is ciga- carcinogen).6 14 A study among firefighters showed high levels of rette smoking,19 which is probably the major source of fluorene phenanthrene (median sum of urinary 1-­OH-Phe,­ 2-­OH-­Phe and exposure among the controls. The mean urinary concentrations 3-­OH-­Phe was up to 3.1 µg/g crea.) and fluorene (∑OH-fluorene­ of ∑OH-­Phe and ∑OH-­Flu in a study among daily smokers who up to 0.81 µg/g crea.).15 High phenanthrene exposure was also smoked ≥5 cigs/day were 2.46 and 1.7 µg/g crea., respectively.20 observed among coke oven workers (median sum of urinary 1-­OH-­ In another study, comparable exposure to fluorene (mean 1.5 µg/L Phe, 2-­OH-­Phe, 3-­OH-­Phe, 4-­OH-­Phe and 9-­OH-­Phe=10.7 µg/g 2-­OH-Flu),­ but not phenanthrene (mean 0.56 µg/L ∑OH-­Phe), crea.), workers in the production of graphite electrodes (5.8 µg/g was reported among daily smokers who smoked ≥5 cigs/day.21 The crea.) and refractory materials (16.8 µg/g crea.).16 Among levels of phenanthrene and fluorene exposure in the latter study workers in electrode production, mean urinary concentrations of are consistent with those observed in our study for the smoking

Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 5 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from

Table 3 Linear regression analyses for the associations between urinary concentrations of ∑OH-fluorene­ (µg/g creatinine) and cancer biomarkers among chimney sweeps, controls and creosote-­exposed workers Model 3 Model 1 Model 2 (age-­adjusted and pack-­years-­ (unadjusted) (age-­adjusted and smoking-­adjusted) adjusted among smokers) B (95% CI) P value B (95% CI) P value B (95% CI) P value Chimney sweeps n=143* n=142* n=27† Relative telomere length −0.013 (−0.036 to 0.0096) 0.25 −0.0046 (−0.031 to 0.022) 0.73 −0.019 (−0.073 to 0.034) 0.46 Relative mtDNAcn −0.0017 (−0.039 to 0.035) 0.93 0.0081 (−0.038 to 0.054) 0.73 0.028 (−0.097 to 0.15) 0.65 F2RL3_CpG1 −6.0 (−7.1 to −4.9) <0.001‡ −4.0 (−5.3 to −2.7) <0.001‡ −5.9 (−9.5 to −2.3) 0.003 F2RL3_CpG2 (cg03636183) −5.1 (−6.2 to −4.0) <0.001‡ −2.7 (−3.9 to −1.5) <0.001‡ −4.2 (−7.6 to −0.8) 0.019 AHRR_CpG1 −10 (−12 to to 8.3) <0.001‡ −5.9 (−8.2 to −3.5) <0.001‡ −8.2 (−13 to −3.5) 0.002 AHRR_CpG2 −10 (−12 to to 8.5) <0.001‡ −5.8 (−7.8 to −3.8) <0.001‡ −7.7 (−12 to −3.7) 0.001 AHRR_CpG3 (cg05575921) −14 (−16 to to 12) <0.001‡ −7.1 (−9.6 to −4.7) <0.001‡ −10 (−16 to to 5.1) 0.001 Controls n=147* n=147* n=25† Relative telomere length −0.016 (−0.044 to 0.012) 0.27 0.049 (−0.0014 to 0.099) 0.06 0.060 (−0.0023 to 0.12) 0.058 Relative mtDNAcn −0.069 (−0.12 to -0.021) 0.005‡ −0.0038 (−0.091 to 0.084) 0.93 −0.012 (−0.10 to 0.079) 0.78 F2RL3_CpG1 −6.5 (−7.8 to −5.1) <0.001‡ −3.9 (−6.3 to −1.6) 0.001‡ −2.4 (−6.7 to 1.9) 0.26 F2RL3_CpG2 (cg03636183) −6.8 (−8.0 to −5.6) <0.001‡ −3.7 (−5.7 to −1.8) <0.001‡ −2.6 (−6.3 to 1.1) 0.15 AHRR_CpG1 −12 (−14 to to 9.9) <0.001‡ −6.8 (−11 to −2.7) 0.001‡ −3.0 (−9.2 to 3.2) 0.32 AHRR_CpG2 −11 (−13 to to 8.9) <0.001‡ −5.7 (-9.3 to -2.0) 0.003‡ −3.3 (−8.9 to 2.3) 0.23 AHRR_CpG3 (cg05575921) −18 (−20 to to 16) <0.001‡ −8.2 (−12 to −4.6) <0.001‡ −5.5 (−11 to 0.1) 0.055 Creosote-exposed­ workers n=19 n=18 n=3 Relative telomere length −0.00045 (−0.0026 to 0.0017) 0.66 −0.001 (−0.0039 to 0.0018) 0.45 – – Relative mtDNAcn 0.00088 (−0.0038 to 0.0055) 0.69 −0.00012 (−0.0066 to 0.0063) 0.97 – – F2RL3_CpG1 −0.10 (−0.24 to 0.037) 0.14 −0.10 (−0.29 to 0.087) 0.27 – – F2RL3_CpG2 (cg03636183) −0.086 (−0.22 to 0.045) 0.18 −0.12 (−0.29 to 0.052) 0.16 – – AHRR_CpG1 −0.081 (−0.36 to 0.20) 0.55 −0.12 (−0.47 to 0.23) 0.47 – – AHRR_CpG2 −0.079 (−0.35 to 0.19) 0.55 −0.12 (−0.46 to 0.22) 0.46 – – AHRR_CpG3 (cg05575921) −0.11 (−0.42 to 0.19) 0.45 −0.062 (−0.40 to 0.27) 0.69 – – Model 1: (cancer biomarker: outcome)=intercept + B.(∑OH-fluorene).­ Model 2: (cancer biomarker: outcome)=intercept + B(∑OH-fluorene)+B1.age+B2.smoking.­ Model 3: (cancer biomarker: outcome)=intercept + B(∑OH-fluorene)+B1.age+B2.(pack-­ ­years) . *There was one missing case for some of the outcomes (eg, AHRR_CpG3_cg05575921).

†There were two missing cases among chimney sweeps and four missing cases among the controls. http://oem.bmj.com/ ‡Significant after adjustment for multiple comparisons (FDR<0.05) B, unstandardised beta estimate; mtDNAcn, mitochondrial DNA copy number; ∑OH-fluorene­ , sum of 2-hydroxyfluorene­ and 3-hydroxyfluorene­ . controls (median ∑OH-­Flu 1.4 µg/g crea.; ∑OH-Phe­ 0.5 µg/g with the smoking controls and significance was reached only for crea.). However, even the non-smoking­ controls showed fluorene chimney sweeps. This indicates that the relationship between fluo- metabolite concentrations in urine, indicating that there are other rene and DNA methylation is not only related to smoking but also

sources of fluorene than smoking in the general environment. It to chimney sweeps’ exposure from work. We could not adjust for on September 27, 2021 by guest. Protected copyright. should be noted that no significant difference was observed for pack-years­ of former smokers because we did not have these data the urinary concentrations of ∑OH-­Flu when comparing smoking for the controls; however, pack-­years of former smokers should controls to smoking chimney sweeps (1.4 vs 1.5 µg/g crea.). These not be correlated with urinary fluorene metabolite concentrations, figures show that cigarette smoking has a greater impact on ∑OH-­ a marker for recent fluorene exposure. It is worth mentioning that Flu levels than chimney sweeping. Therefore, the question arose our analysis would have benefited from adjustment for blood cell whether the association between fluorene and DNA hypomethyla- composition, however, we do not expect systematic variations in tion is due to cigarette smoking only or does fluorene occupational blood cell composition between study groups because of the simi- exposure play a role? larities in age, health status (healthy workers) and smoking status. We evaluated this question in different ways. We started by The inverse association between fluorene and DNA methylation adjusting for age and smoking in the linear regression models; of F2RL3 and AHRR was corroborated by the mediation analyses however, this adjustment did not seem to fully eradicate the effect where fluorene metabolite concentrations showed mediation effect of smoking on the association between fluorene exposure and for the associations between chimney sweeping (being a chimney DNA methylation, as seen in the models for the controls (table 3; sweep) versus DNA hypomethylation. Nevertheless, these findings model 2). This may be caused by the residual confounding of should still be cautiously interpreted, as in the analysis of never smoking, which in this case can be due to differences in pack-years­ smoking chimney sweeps and controls, only AHRR CpG2 showed in smokers and former smokers. We, therefore, restricted the anal- significantly lower methylation with increasing ∑OH-­Flu in the ysis to smoking chimney sweeps and smoking controls adjusting sweeps. for age and pack-years.­ We found that the negative effect estimates Even though creosote-­exposed workers were highly exposed of fluorene metabolite concentrations on DNA methylation were to fluorene, they did not show any association between fluorene twice as pronounced among smoking chimney sweeps compared exposure and DNA methylation. Likely, the small sample size

6 Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from

limited our possibilities to detect small to moderate effects. But there may also be other possibilities for the lack of associations. Fluorene might be a proxy for other toxicant(s) present in the work environment of the chimney sweeps (eg, black carbon particles), Mediation% , sum of , but not in the work environment of the creosote-­exposed workers. ­ Phe In both occupations, exposure to PAHs occurs through inhalation ­ OH- and dermal contact, but the predominant route of exposure to certain PAH might differ.5 22–24 Inhalation of particles could lead ∑2-,3- to inflammatory responses that may cause the hypomethylation CI) of the CpG cites examined in this study by yet unknown mecha-

AHRR_ cg05575921 B (95% −2.0 (−4.2 to 0.063) −2.2 (−4.3 to 0.038) -6 −2.1 (−4.4 to 0.18) -3 −2.0 (−4.2 to 0.19) 1 −2.1 (−4.4 to 0.15) -3 −0.64 (−2.6 to 1.3) 69 nisms. Indeed, exposure to diesel exhaust has been linked to DNA alterations of CpG sites related to genes involved in inflammation and oxidative stress response.25 Chimney sweeps that are mainly hydroxyphenanthrene; ­ exposed to PAHs from soot particles through inhalation could, therefore, show associations between DNA methylation levels and Mediation%

, and 4- , fluorene exposure. Creosote-­exposed workers, on the other hand, are predominantly dermally exposed to non-­particulate PAHs from creosote oil5 24, and thus no such association could be found. In addition, the PAH content varies between soot and creosote oil. For instance, BaP metabolites were found in chimney sweeps’ CI)

urine, but not in creosote-exposed­ workers’ urine.8 Still, BaP did ­ hydroxyphenanthrene AHRR _CpG2 B (95% −2.6 (−4.4 to −0.69) −2.7 (−4.6 to −0.76) -6 −2.7 (−4.7 to −0.72) -7 −2.6 (−4.5 to −0.64) -1 −2.7 (−4.7 to −0.72) -6 −1.5 (−3.3 to 0.31) 42 not show any association with DNA methylation in our study. It , 3- , is also possible that there is no relationship between fluorene and DNA methylation at such extreme levels of fluorene exposure among creosote-­exposed workers or that the association is non-­ linear above a certain level of exposure. A study on exposure from Mediation% cigarette smoking including current smokers (n=172) found that ­ hydroxyphenanthrene participants in the second quartiles Q2 (27 cigs/day), Q3 (40 cigs/

, 2- , day) and Q4 (53 cigs/day) had comparable hypomethylation of AHRR cg05575921 compared with Q1 (19 cigs/day).26 Likewise, CI)

when examining the cumulative exposure (pack-years),­ the authors found that participants in Q2 (23 pack-years),­ Q3 (38 pack-years)­ AHRR _CpG1 B (95% −1.5 (−3.6 to 0.54) −1.7 (−3.8 to 0.49) -8 −1.7 (−4.0 to 0.52) −12 −1.7 (−3.8 to 0.49) -8 −1.7 (−3.9 to 0.52) −10 −0.40 (−2.4 to 1.6) 74 and Q4 (57 pack-years)­ had similar AHRR cg05575921 methyla- tion, and concluded that exposure from current smoking impacts ­ hydroxyphenanthrene DNA methylation, however, very high levels of smoking exposure adds only a negligible effect.26 http://oem.bmj.com/ , sum of 1- , – – – – – Mediation% Despite the frequent presence of fluorene in PAH mixtures, only

Phe ­ a limited number of studies have explored its toxicity. An in vitro study using Chinese hamster lung cell line reported a clastogenic

∑OH- effect of fluorene only in the presence of metabolic activation (rat S9 mix).27 Fluorene also showed axial development toxicity DNA hypomethylation), where being a chimney sweep is a binary variable (chimney sweep vs control), PAH PAH (chimney sweep vs control), where being a chimney sweep is binary variable exposure → DNA hypomethylation), PAH CI) in zebrafish embryos and sea urchin embryos through dysregula-

L3_cg03636183* 28 29 R tion of the β-catenin/Wnt signalling pathway. An in ovo study F2 B (95% 0.02 (−1.02 to 1.07) – – – – – – on September 27, 2021 by guest. Protected copyright. hydroxyfluorene; ­ assessing histopathological changes in the chicken fetal liver on exposure to various compounds found no genotoxicity, measured as DNA adducts, of fluorene.30 31 Due to the scarcity of toxicolog- ical studies, and thus, the inadequate evidence of fluorene carcino- 6

Mediation% genicity, IARC has classified fluorene in group 3. It is biologically plausible that PAHs could induce alterations in DNA methylation. ­ hydroxyfluorene and 3- The key metabolic pathway for PAHs is the oxidisation to reactive metabolites (eg, epoxides: phase I) followed by conjugation (eg, 32

.exposure group (chimney sweep vs control)+B1.age+B2.smoking (three categories). metabolite). .exposure group (chimney sweep vs control)+B1.age+B2.smoking (three categories)+B3.(PAH glutathione: phase II). As a result of the phase II metabolism, B B sum of 2- _CpG1 CI) PAH exposure could lead to elevated levels of glutathione whose Flu, ­ biological synthesis is dependent on homocysteine.33 This process F2RL3 B (95% −0.98 (−2.2 to 0.22) −0.73 (−2.0 to 0.56) 26 −0.77 (−2.05 to 0.51) 21 −0.30 (−1.4 to 0.84) 69 −0.91 (−2.2 to 0.34) 7 −0.79 (−2.0 to 0.46) 20 may result in reduced levels of S-adenosylmethionine­ (SAM), a key ­ hydroxyphenanthrene. methyl group donor for DNA methylation, and subsequently lead to DNA hypomethylation.33

Conclusions Mediation analysis for the path (being a chimney sweeps →

­ Phe Chimney sweeps and creosote-­exposed workers are occupationally

­ OH-

­ Phe ­ Phe exposed to phenanthrene and fluorene. We report for the first time ­ ­ Phe Flu ­ hydroxyphenanthrene and 3- ­ ­ OH- OH- that fluorene exposure is associated with the DNA hypomethyla- ase model† 2- exposure is a continuous variable (PAH metabolites) and DNA hypomethylation is a continuous variable (methylation of F2RL3 and AHRR ) metabolites) and DNA hypomethylation is a continuous variable (PAH exposure is a continuous variable (µg/g crea‡ B (µg/g crea.)‡ (µg/g crea.)‡ no association with exposure group (p=0.96). not performed for F2RL3 _cg03636183 because there was *Mediation analysis was †Outcome (DNA methylation)=intercept + (µg/g crea.)‡ (ng/g crea.)‡ ‡Outcome (DNA methylation)=intercept + ∑OH- unstandardised beta estimate; B, 1- ∑2-,3- 4- ∑OH- ∑OH- Table 4 Table tion of F2RL3 and AHRR, prospective markers for lung cancer, in

Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413 7 Exposure assessment Occup Environ Med: first published as 10.1136/oemed-2020-106413 on 8 May 2020. Downloaded from the chimney sweeps. Further studies are needed to clarify whether 7 Persson B, Dahlander AM, Fredriksson M, et al. Malignant lymphomas and fluorene or other PAHs/toxicants are underlying the epigenetic occupational exposures. Br J Ind Med 1989;46:516–20. 8 alhamdow A, Lindh C, Hagberg J, et al. Dna methylation of the cancer-­related genes alterations observed in workers exposed to PAHs. F2RL3 and AHRR is associated with occupational exposure to polycyclic aromatic hydrocarbons. Carcinogenesis 2018;39:869–78. Author affiliations 9 Fasanelli F, Baglietto L, Ponzi E, et al. Hypomethylation of smoking-related­ genes 1 Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden is associated with future lung cancer in four prospective cohorts. Nat Commun 2 Department of Laboratory Medicine, Division of Occupational and Environmental 2015;6:10192. Medicine, Lund University, Lund, Sweden 10 Barthel FP, Wei W, Tang M, et al. Systematic analysis of telomere length and somatic 3 Region Stockholm, Centre for Occupational and Environmental Medicine, alterations in 31 cancer types. Nat Genet 2017;49:349–57. Stockholm, Sweden 11 hu L, Yao X, Shen Y. Altered mitochondrial DNA copy number contributes to human 4 School of Public Health and Community Medicine, Section of Occupational and cancer risk: evidence from an updated meta-­analysis. Sci Rep 2016;6:35859. Environmental Medicine, University of Gothenburg Sahlgrenska Academy, Göteborg, 12 Baglietto L, Ponzi E, Haycock P, et al. Dna methylation changes measured in pre-­ Sweden diagnostic peripheral blood samples are associated with smoking and lung cancer 5 MTM Research Centre, Örebro universitet Akademin för Naturvetenskap och Teknik, risk. Int J Cancer 2017;140:50–61. Orebro, Sweden 13 gliga AR, Engström K, Kippler M, et al. Prenatal arsenic exposure is associated with 6 Department of Occupational and Environmental Medicine, Faculty of Medicine and increased plasma IGFBP3 concentrations in 9-­year-­old children partly via changes in Health, Örebro University, Örebro, Sweden DNA methylation. Arch Toxicol 2018;92:2487–500. 7 Department of Chemical and Biological Work Environment, STAMI, Oslo, Norway 14 Jongeneelen FJ. Benchmark guideline for urinary 1-­hydroxypyrene as biomarker of occupational exposure to polycyclic aromatic hydrocarbons. Ann Occup Hyg Acknowledgements The authors would like to thank all study participants, 2001;45:3–13. the occupational nurses Pia Tallving, Patrice Milton, Eva Assarsson, Göte Mölleby 15 Fent KW, Toennis C, Sammons D, et al. Firefighters’ absorption of AHP s and VOCs and the biomedical analyst Ina Lindell for recruiting the workers in the study. They during controlled residential fires by job assignment and fire attack tactic. J Expo Sci also thank the chimney sweeps’ trade union (Kommunal), and chimney sweeps’ Environ Epidemiol 2020;30:338–49. employer organisation (Sveriges Skorstensfejaremästares Riksförbund) for their 16 rossbach B, Preuss R, Letzel S, et al. Biological monitoring of occupational exposure cooperation. to polycyclic aromatic hydrocarbons (PAH) by determination of monohydroxylated metabolites of phenanthrene and pyrene in urine. Int Arch Occup Environ Health Contributors AA: drafted the manuscript and performed the statistical analysis. 2007;81:221–9. YJE: drafted the manuscript and performed the PAH metabolites analysis. AMK and 17 Barbeau D, Lutier S, Marques M, et al. Comparison of gaseous polycyclic aromatic CHL: planned and optimised the PAH metabolites analysis. PG and MA: interpreted hydrocarbon metabolites according to their specificity as biomarkers of occupational the study findings.H T: recruited chimney sweeps and creosote-­exposed workers and exposure: selection of 2-hydro­ xyfluorene and 2-hydro­ xyphenanthrene. J Hazard Mater carried out the occupational hygienic assessment. JH and PG: recruited creosote-­ 2017;332:185–94. exposed workers. KB: designed the study, recruited the chimney sweeps and controls, 18 gündel J, Schaller KH, Angerer J. Occupational exposure to polycyclic aromatic and interpreted the data. All authors critically reviewed the manuscript and agreed hydrocarbons in a fireproof stone producing plant: biological monitoring of with the final version. 1-hydro­ xypyrene, 1-, 2-, 3- and 4-hydro­ xyphenanthrene, 3-hydro­ xybenz(a)anthracene and Funding This study was financially supported by the Swedish Research Council 3-­hydroxybenzo(a)pyrene. Int Arch Occup Environ Health 2000;73:270–4. for Health, Working Life and Welfare (FORTE) (grant#: 2012-00402), the Medical 19 Vu AT, Taylor KM, Holman MR, et al. Polycyclic aromatic hydrocarbons in the Training and Research Agreement (ALF grants; Region Örebro län) (grant#: OLL- mainstream smoke of popular U.S. cigarettes. Chem Res Toxicol 2015;28:1616–26. 550721), AFA Insurance (AFA Försäkring) (grant#: 120115) and Karolinska Institutet. 20 goniewicz ML, Gawron M, Smith DM, et al. Exposure to nicotine and selected YJE was supported by a fellowship from the German Research Foundation (DFG). toxicants in cigarette smokers who switched to electronic cigarettes: a longitudinal Within-­Subjects observational study. Nicotine Tob Res 2017;19:160–7. Competing interests None declared. 21 ramsauer B, Sterz K, Hagedorn H-­W, et al. A liquid chromatography/tandem mass Patient consent for publication Not required. spectrometry (LC-­MS/MS) method for the determination of phenolic polycyclic aromatic hydrocarbons (OH-­PAH) in urine of non-­smokers and smokers. Anal Bioanal Provenance and peer review Not commissioned; externally peer reviewed. Chem 2011;399:877–89. http://oem.bmj.com/ Data availability statement All data relevant to the study are included in the 22 aTSDR. Toxicological profile for wood creosote, coal TAR creosote, coal TAR, coal TAR article or uploaded as supplementary information. pitch, and coal TAR pitch volatiles. Atlanta, GA, USA: Agency for Toxic Substances and Open access This is an open access article distributed in accordance with the Disease Registry, US Department of Health and Human Services, Public Health Service, Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits 2002. others to copy, redistribute, remix, transform and build upon this work for any 23 Kammer R, Tinnerberg H, Eriksson K. Evaluation of a tape-­stripping technique purpose, provided the original work is properly cited, a link to the licence is given, for measuring dermal exposure to pyrene and benzo(a)pyrene. J Environ Monit and indication of whether changes were made. See: https://​creativecommons.​org/​ 2011;13:2165–71. licenses/by/​ ​4.0/.​ 24 heikkilä P. Respiratory and dermal exposure to creosote. Kuopio, Finland: Kuopio University Printing Office, 2001. on September 27, 2021 by guest. Protected copyright. ORCID iDs 25 Ferrari L, Carugno M, Bollati V. Particulate matter exposure shapes DNA methylation Ayman Alhamdow http://orcid.​ ​org/0000-​ ​0002-3643-​ ​0518 through the lifespan. Clin Epigenetics 2019;11:129. Per Gustavsson http://orcid.​ ​org/0000-​ ​0003-2221-​ ​8599 26 su D, Wang X, Campbell MR, et al. Distinct epigenetic effects of tobacco smoking in Pål Graff http://orcid.​ ​org/0000-​ ​0003-4928-​ ​617X whole blood and among leukocyte subtypes. PLoS One 2016;11:22. Karin Broberg http://orcid.​ ​org/0000-​ ​0002-5862-​ ​468X 27 Matsuoka A, Sofuni T, Miyata N, et al. Clastogenicity of 1-nitropyrene­ , dinitropyrenes, fluorene and mononitrofluorenes in cultured Chinese hamster cells. Mutat Res 1991;259:103–10. References 28 Fairbairn EA, Bonthius J, Cherr GN. Polycyclic aromatic hydrocarbons and dibutyl 1 Pott P. Chirurgical Observations Relative to the Cataract, the Polypus of the Nose, the phthalate disrupt dorsal–ventral axis determination via the Wnt/β-catenin signaling Cancer of the Scrotum, the Different Kinds of Ruptures, and the Mortification of the pathway in zebrafish embryos. Aquat Toxicol 2012;125:188–96. Toes and Feet. In: Carnegy TJ, Hawes L, Clarke W, et al, eds. London, 1775: 291–5. 29 Pillai MC, Vines CA, Wikramanayake AH, et al. Polycyclic aromatic hydrocarbons 2 Pukkala E, Martinsen JI, Lynge E, et al. Occupation and cancer - follow-up­ of 15 disrupt axial development in sea urchin embryos through a β-catenin dependent million people in five Nordic countries. Acta Oncol 2009;48:646–790. pathway. Toxicology 2003;186:93–108. 3 hogstedt C, Jansson C, Hugosson M, et al. Cancer incidence in a cohort of Swedish 30 iatropoulos MJ, Kobets T, Duan J-­D, et al. Chicken egg fetal liver DNA and chimney sweeps, 1958-2006. Am J Public Health 2013;103:1708–14. histopathologic effects of structurally diverse carcinogens and non-­carcinogens. Exp 4 alhamdow A, Lindh C, Albin M, et al. Early markers of cardiovascular disease are Toxicol Pathol 2017;69:533–46. associated with occupational exposure to polycyclic aromatic hydrocarbons. Sci Rep 31 Williams GM, Duan J-­D, Brunnemann KD, et al. Chicken fetal liver DNA damage and 2017;7:9426. adduct formation by activation-­dependent DNA-­reactive carcinogens and related 5 iarc. Chemical agents and related occupations: a review of human carcinogens. soot, compounds of several structural classes. Toxicol Sci 2014;141:18–28. as found in occupational exposure of chimney sweeps. Lyon, France: International 32 Moorthy B, Chu C, Carlin DJ. Polycyclic aromatic hydrocarbons: from metabolism to Agency for Research on Cancer, IARC Press, 2012: 100F. 209–14. lung cancer. Toxicol Sci 2015;145:5–15. 6 iarc. Monographs on the evaluation of carcinogenic risks to humans. Some non-­ 33 lee D-­H, Jacobs DR, Porta M. Hypothesis: a unifying mechanism for nutrition and heterocyclic polycyclic aromatic hydrocarbons and some related exposures. Lyon, chemicals as lifelong modulators of DNA hypomethylation. Environ Health Perspect France: International Agency for Research on Cancer, IARC Press, 2010: 92. 754–73. 2009;117:1799–802.

8 Alhamdow A, et al. Occup Environ Med 2020;0:1–8. doi:10.1136/oemed-2020-106413